Munich Personal RePEc Archive
Macroeconomic instability in
Afghanistan: causes and solutions
Joya, Omar
University of Bordeaux
September 2011
Online at https://mpra.ub.uni-muenchen.de/37658/
MPRA Paper No. 37658, posted 26 Mar 2012 16:05 UTC
Master’s dissertation ( )
Master II “Applied Economics: Development”
Macroeconomic Instability in Afghanistan:
Causes and Solutions
by Omar Joya
Supervisor of research:
Date of academic defence:
This dissertation contributes to an increasing literature on macroeconomic instability in developing countries. It makes a critical review of the literature and classifies the sources of instability under exogenous and endogenous factors. It then argues that the impact of exogenous shocks is determined by the structural characteristics of the economy which act as a risk*management mechanism. The paper also explains that macroeconomic instability is both a cause and a reflection of underdevelopment. Whilst macroeconomic instability constraints the long*term growth and thus development, it is also the result of the co*
existence of various ‘underdeveloped structures’ in the economy. The paper also presents a case study on Afghanistan. Through a diagnostic approach, it identifies the sources of instability in the country and proposes a series of policies and reforms in order to overcome macroeconomic instability in Afghanistan.
Le présent mémoire est une contribution aux études sur l’instabilité macroéconomique dans les pays en développement. Il fait un compte*rendu de la littérature, et classifie les sources de l’instabilité macroéconomique sous des facteurs exogènes et endogènes. Ensuite, il soutient que l’effet des chocs exogènes est déterminé par les caractéristiques structurelles de l’économie qui fonctionnent comme un mécanisme de gestion de risque. Ce mémoire explique que l’instabilité macroéconomique est à la fois une cause et un reflet de sous*
développement. D’une part, l’instabilité macroéconomique contraint la croissance de long*
terme et donc le développement, d’autre part, l’instabilité est le résultat de coexistence de différentes structures sous*développées dans l’économie. Ce mémoire présente aussi un cas d’étude sur l’Afghanistan. A travers d’une approche diagnostique à ce sujet, il identifie les sources de l’instabilité dans le pays et propose une série des politiques et des réformes en vue de surmonter l’instabilité macroéconomique en Afghanistan.
JEL classification: E32, E60, O11, O53
Keywords: Macroeconomic stability, Macroeconomic volatility, Macroeconomic instability, Developing countries, Afghanistan, Diagnostic approach, Policy analysis
Author’s email address: omr.joya@gmail.com
Introducing key questions ... 1
... 4
1. Definition and some stylised facts ... 4
1.1. Macroeconomic stability... 4
1.2. Macroeconomic volatility ... 5
1.3. Fluctuations in Real Business Cycle theory... 7
1.4. Measuring volatility ... 8
1.5. Stylised facts ... 11
2. Costs of macroeconomic instability ... 15
2.1. Negative impact on investment and long*term growth ... 15
2.2. Increase in inequality and poverty... 17
2.3. Welfare costs ... 19
3. Sources of macroeconomic instability ... 21
3.1. Exogenous factors ... 21
... 21
! ... 23
" ! ... 23
# ... 25
# $ # ! ... 26
3.2. Endogenous factors ... 26
% ... 27
& ... 29
' # ... 30
# ' ! ! ! ... 31
# ( ... 32
# ) ... 34
4. Macroeconomic instability and development ... 35
! ... 38
1. The Afghan economy ... 38
1.1. Recent economic history ... 38
1.2. The economy since 2002 ... 40
1.3. Foreign aid in Afghan economy ... 48
1.4. Macroeconomic volatility in Afghanistan ... 51
2. Diagnostic approach to macroeconomic instability in Afghanistan ... 53
2.1. Sources of macroeconomic instability ... 54
* # ! ... 54
" + # ! ... 59
,# ! ... 61
2.2. Solutions ... 66
- !. ... 67
A. Agriculture sector ... 67
B. Natural resources ... 70
' # ! ... 76
$ / / ! ... 79
# ' # ... 80
Conclusion ... 81
References ... 84
Af./Afs. Afghani(s)
AISA Afghanistan Investment Support Agency CSO Central Statistics Office, Afghanistan
CPI Consumer Prices Index
DAB Da Afghanistan Bank (Central Bank)
DAC Development Co*operation Directorate, OECD
DFID Department for International Development, United Kingdom
IMF International Monetary Fund
FAO Food and Agriculture Organization
MAIL Ministry of Agriculture, Irrigation and Livestock, Afghanistan MoCI Ministry of Commerce and Industries, Afghanistan
MoF Ministry of Finance, Afghanistan NATO North Atlantic Treaty Organization
NRVA National Risk and Vulnerability Assessment ODA Official development assistance
OECD Organisation for Economic Co*operation and Development R&D Research and Development
SH Solar Hijri (Persian calendar year)
UNCTAD United Nations Conference on Trade and Development UNODC United Nations Office on Drugs and Crime
UNSD United Nations Statistics Division USGS United States Geological Survey
WDI World Development Indicators, World Bank
WEO World Economic Outlook, IMF
WTO World Trade Organization
"
The official calendar in Afghanistan is the Persian calendar, known as 0 1 . Its years are designated $* (Anno Persico) or 0. Financial year in Afghanistan is also adjusted according to the SH year, which starts on March 21st in the Gregorian calendar. For example, the SH 1389 corresponds to Mar 21, 2010 – Mar 20, 2011. The annual national accounts data on Afghanistan is usually calculated over the SH years, regardless of the source of data. For example, the data on GDP whether reported by foreign sources such as IMF or UNSD or by local sources such as CSO or DAB refer to SH years. For simplicity, the annual national accounts data are sometimes indicated in a single Gregorian format such as 2010, instead of 2010/11 (which both refer to 1389).
A billion means a thousand million (= 109) and a trillion means a million million (= 1012).
" #
Developing countries have always been characterized with economic volatility and an uncertain macroeconomic environment. While developed countries have enjoyed stability since the 1980s, macroeconomic instability has been a serious concern in the developing world. From the Latin American debt crisis in 1982, to the Asian financial crisis in 1997, and to the world food price crisis in 2007, developing countries have suffered from serious volatilities in output growth, inflation, exchange rate, interest rates, and other variables of concern. These macroeconomic volatilities are not only observed in low*income countries (LICs), but they are also present in middle*income economies. However, the source and nature of these volatilities differ from one group to another. The magnitude, depth and persistence of macroeconomic volatility are more pronounced in poor and least*developed countries (LDCs) than in more developed ones. For low*income countries, macroeconomic instability is of a major concern because it seriously affects the poor and has negative impact on their long*term growth.
In a seminal paper, Lucas (1988) attracted the attention of economists to this phenomenon, noting that “within the advanced countries, growth rates tend to be very stable over long periods of time.... For poorer countries, however, there are many examples of sudden, large changes in growth rates, both up and down.” Since then, economists have specifically been interested in studying macroeconomic instability in developing countries.
However, in traditional macroeconomics, there tended to be a dichotomy between “growth”
and “volatility” in economic aggregates. Growth theory and Real Business Cycle (RBC) theory have traditionally been treated as unrelated areas of macroeconomics. Therefore, for a long time, economic volatility was treated as a secondary phenomenon in the business cycle literature. It was considered a second*order issue of interest and a phenomenon related to the fluctuations in the business cycle.
However, since the seminal paper of Ramey and Ramey (1995) which showed that volatility and growth rate are strongly correlated, research on macroeconomic volatility has been carried out with methods and models independent from the RBC theory.
Macroeconomic instability has thus developed into its own field of research over the last decade, thanks to the recognition that “non*linearities ... magnify the negative effects of volatility on long*run growth and inequality, especially in poor countries” (Aizenman and Pinto, 2005).
The concept of macroeconomic stability/instability was popularised during the 1980s with the stabilisation policies prescribed by the Washington*based institutions (i.e.
International Monetary Fund and the World Bank) to the developing countries affected by the debt crisis. However, macroeconomic stability should not be conceived exclusively in the context of stabilisation policies of the IMF. Under the Structural Adjustment Programs (SAPs), macroeconomic stability was defined in a very narrow sense; focusing primarily at low inflation, price stability, and low fiscal and current*account deficits. Nonetheless, such a narrow consideration was criticized by some economists from the academic milieu, both for not considering other important variables (mainly real variables, including unemployment) and for considering a very narrow margin of variability for some variables; for example insisting on single digit threshold for inflation (Stiglitz et al., 2006). The concept of macroeconomic stability has undergone considerable changes in the economic discourse. The contemporary definition of macroeconomic stability enjoys a much broader sense.
It should also be noted that the concept of macroeconomic volatility is not necessarily associated with economic crises. Although volatility usually appears during the periods of crisis in developed countries, it is an endemic phenomenon in developing countries and must not be confined to instances of crisis1 (Malik and Temple, 2009). Moreover, a period of macroeconomic volatility is not necessarily a period of recession. A country can well suffer from macroeconomic volatility without “formally” being into an economic recession.2
This dissertation contributes to the literature on macroeconomic instability in developing countries. The first part of this dissertation seeks to answer three sets of questions.
First, what are the possible costs of macroeconomic volatility in terms of welfare and other economic indicators? Are there costs associated with macroeconomic instability or is macro instability neutral in regard to the welfare of the economy? Secondly, this dissertation identifies the exogenous and endogenous sources of macroeconomic instability by reviewing the results of empirical and theoretical studies. Finally, this paper seeks to answer if macroeconomic instability is a cause a reflection of underdevelopment. This question is crucial for policy analysis, because if macroeconomic instability is a source of underdevelopment, then overcoming instability would be a key to prosperity and a solution to all underlying problems in LICs. And if it is a consequence and a reflection of underdevelopment, then instead of focusing on policies to overcome instability, policy*
makers should engage with broad*based structural and development policies. But if it is both
1 For example, recent instances of crisis in the developing world include Mexico in 1995, Russia in 1998, Turkey in 2001, Argentina in 2002, and world commodity prices crisis in 2008.
2 Although there is no formal definition for “economic recession,” but as a rule of thumb ‘two consecutive quarters of a decline in real GDP’ is considered a recession.
a cause and a reflection of underdevelopment, then it requires a more complex and in*depth analysis of the situation.
The second part of this dissertation presents a case study on Afghanistan and makes diagnosis of macroeconomic instability in the country. It identifies the sources of instability and suggests a series of policies and reforms to overcome and to correct instability in the country. Afghanistan can be a good example for the analysis of macroeconomic instability, because since its political shift in 2002 it has experienced serious oscillations in economic growth and price level. The limitation of this paper is that it does not take into account the post*conflict explanations in analysing macroeconomic instability in Afghanistan. The analytical framework presented for the analysis of instability ignores the post*conflict characteristics of the country.
The methodology employed in the two parts of this dissertation is different. In Part I, I make a critical review of the literature on macroeconomic instability in developing countries, and I classify the sources of instability under exogenous and endogenous factors. By classifying them so, I will show that these are the structural characteristics of the economy which determine the nature and level of impact of exogenous shocks on the economy. In Part II, I employ a diagnostic approach to treat macroeconomic instability in Afghanistan. First, I will identify the sources of instability in Afghanistan through quantitative and qualitative analytical methods. Secondly, I will propose some general policies which can help reduce the economy’s exposure to external shocks and install stability in the macroeconomic environment.
The first part of this dissertation is divided into four sections. The first section attempts to give a definition of macroeconomic instability and presents some stylised facts on macroeconomic volatility. The second and third sections make review of the literature on macroeconomic instability; they explain economic costs of macroeconomic volatility and identify the exogenous and endogenous factors which induce instability in the economy.
Finally, the last section explains whether macroeconomic instability is a source or a reflection of underdevelopment.
$ %
$ $
The concept of macroeconomic stability is widely used in the policy*oriented literature, but is almost never properly defined. Based on a large literature which deals with this subject but which has rarely attempted to formally define this term, I present a definition which covers the various – and yet closely related – meanings understood from this concept.
Macroeconomic stability can be described as a situation in which: (i) the level and growth in key macroeconomic variables, as well as the relevant balances between them, are sustainable;
(ii) variability of macroeconomic variables is moderate and lies within an acceptable range;
and/or (iii) full uncertainty regarding the macroeconomic environment does not exist.
The first part of the definition refers to having a sustainable growth rate, low unemployment, moderate inflation, and enjoying internal and external balances; for example, balance between domestic demand and output, balance of payments, fiscal balance, and balance between savings and investment. However, these relationships need not necessarily be in exact balance (IMF, 2001). Imbalances such as fiscal and current account deficits or surpluses can perfectly exist in a stable macroeconomic environment, provided if they are . Furthermore, there is no unique set of thresholds for each macroeconomic variable between stability and instability (IMF, 2001), and there is no consensus on the range within which the levels of these variables should lie. For example, the IMF strongly emphasizes on keeping inflation rate in single digits or even as lower as possible, while, on the other hand, other economists maintain that having an inflation rate between 20 and 30 percent is totally sustainable for developing countries and will not have any negative effect on their growth (Stiglitz et al., 2006).
The second part of the definition implies that the variability of macroeconomic variables should be small. But defining a range for each variable would be inaccurate and improper, because the amplitude of fluctuation for a given variable would depend on the level of balance between other relevant macroeconomic identities. However, the exchange rate has sometimes been subjected to the establishment of a ‘range’ under monetary management systems and stability pacts. For example, the Bretton Woods agreement initially set a one*
percent band for the pegged exchange rates vis*à*vis the US dollar, and the Maastricht criteria fixed the exchange rate fluctuation for the members of the Economic and Monetary Union of the European Union at a margin of 2.25 percent.
The third component reflects the idea that the behaviour and overall movement of macroeconomic variables should be predictable by economic agents and should not subject to full uncertainty. For example, an environment where investors can predict the future rates of growth and inflation and where there is no major uncertainty over the policy makers’
decisions can be characterized with macroeconomic stability.
Historically, during the post*war years dominated by Keynesian thinking, macroeconomic stability basically meant a mix of external and internal balance, which in turn implied full employment and stable economic growth, accompanied by low inflation. During the 1970s and 1980s (and further during 1990s), price stability, and fiscal and current*account balances moved to the centre of attention, supplanting the Keynesian emphasis on “real”
economic activity. In recent years, the emphasis has once again been put on stability (unemployment re*gaining importance), long*term sustainable and / growth, and healthy financial sector (Ocampo, 2005). Stiglitz et al. (2006) emphasize that focus should not only be on price stability but on real variables (real output, unemployment, and inequality) as well, and one has to distinguish between intermediate goals (such as inflation) and final objectives (such long*term, equitable growth).
$ &
Despite the fact that ‘macroeconomic instability’ and ‘macroeconomic volatility’ tend to be employed interchangeably and are closely inter*related, there exists, however, a minor difference between these two terms.
Montiel and Servén (2004) refer macroeconomic instability to “phenomena that decrease the predictability of the domestic macroeconomic environment.” Some other economists, however, define macroeconomic instability in a much broader sense, as “a situation of economic malaise, where the economy does not seem to have settled in a steady
position” (Azam, 2001). Macroeconomic instability can take the form of # ! of key macroeconomic variables or of ! in their behaviour (Montiel and Servén, 2004). Thus, in addition to the concept of volatility, “unsustainable” performances in macroeconomic variables (such as low and unstable growth rate, high inflation, large unemployment, unsustainable fiscal and current*account deficits, etc.) are also included in the definition of instability, while macroeconomic volatility refers uniquely to large fluctuations in macro variables and to the uncertainty associated with them. There can well be a situation which could qualify as of macroeconomic instability, but not as macroeconomic volatility;
for example, a country which suffers from low economic growth, high inflation and large fiscal deficit, but their respective rates and levels are stable and non*volatile.
Hence, this paper defines macroeconomic instability as a situation where: (i) unsustainable imbalances appear in the economy; (ii) variability in key macroeconomic variables is large (i.e. exceeding a certain threshold); and/or (iii) macroeconomic environment is highly uncertain.
It would not be irrelevant to elaborate the differences between volatility, uncertainty, and risk. Aizenman and Pinto (2005) make the following distinction between the three of them: 2 ! describes a situation where several possible outcomes are associated with an event, but the assignment of probabilities to the outcomes is not possible. , in contrast, permits the assignment of probabilities to the different outcomes. 3 ! 4 # ! 4 is allied to risk in that it provides a measure of the possible variation or movement in a particular economic variable or some function of that variable, such as growth rate. It is measured based on observed realizations of a random variable over some historical period.
Conceptually, total variability can be decomposed into ‘predictable’ and ‘unpredictable’
components. Unpredictable variability captures pure risk or uncertainty, and constitutes a
“ .” It can be measured or computed as the residual from a forecasting equation for total variability.
Another distinction is sometimes made between “ # !. and “
# !.” Crisis volatility is a continuum of large or extreme shocks, exceeding a certain cut*off point. There are three methods to define extreme volatility: the imposition of an absolute threshold in magnitude (for example, commodity price changes of more than 10 percent), the imposition of a distributional threshold (the 5 percent largest declines), and the use of a deviation criterion (observations that are at least 2 standard deviations above the mean) (Wolf, 2005). It is also important to note that risks associated with macroeconomic volatilities are which affect most or all economic sectors equally,
in contrast to ! which affect only specific individuals or particular groups of economic sectors (World Bank, 2000).
In this dissertation, the terms volatility and instability may have been used synonymously in some places.
$ ' ( !
Traditional macroeconomic theory suggested that transitory shocks do not have irreversible and permanent effects. Therefore, analysis of fluctuations was done in the context of aggregate supply/aggregate demand model, while evolution of long term variables was analysed through growth models. This dichotomy between the theoretical analysis of fluctuations and of growth relates to the static decomposition between ! and 5 it therefore assumes that shocks do not have permanent effect on the level of a series.
However, the dichotomy between cycle and trend was challenged by several empirical and theoretical researches during the 1970s and 1980s. These studies showed that short*term movements in all macroeconomic aggregates have an impact on the long*run level of their series (i.e. their trends). In other words, transitory shocks which are at the basis of cyclical phenomenon persist in the long run. Macroeconomic time series are, thus, composed of permanent (trend) and cyclical components. However, the acknowledgement of this fact has serious implications. At the statistical level, it makes the traditional dichotomy between cycle and trend unmeaningful. In fact, the trend cannot be considered independent of and unaffected from transitory shocks. And at the theoretical sphere, it requires analysing the fluctuations and the growth in a unified way (Hairault, 2000). This latest methodology
constitutes the principals of the ! 67 !.
The RBC model extends the Neo*classical growth model in three main ways: First, it adds a which allows for the possibility of variable employment over time, and thus flexible wages. The RBC theory further assumes that prices in other markets are also flexible and that markets always clear out. Secondly, it allows for to exogenous real variables. In particular, it allows for variations in “technology” and/or government spending. As a result, households and firms face uncertainty regarding future variables. Finally, it assumes that economic agents make about the future and operate in competitive markets.
In general, RBC theory models the economy using dynamic general equilibrium models (DGEM). A simple RBC model is based on the same aggregate function as that in a neoclassical growth model with constant return to scales:
( , ) 8 =% 9 $ :
(1)
where $ is an exogenous process of technology which evolves according to a
!model, such as:
ln$ =ln$ + +;
(2)
ln$ is a constant, is the trend growth rate (assumed to be known with certainty) and ; represents deviations around the trend. These deviations from trend are further assumed to follow a first*order autoregressive process:
1
; =ρ;; − +ε (3)
where ρ;is a persistence parameter and ε represents a “technology shock.”
Hence, according to RBC theory, shocks which induce fluctuations and cyclical behaviour are induced by # ! and these technological and productivity shocks are persistent over some period of time (depending on the value of ρ; ).
Movements in output and employment are thus seen as efficient responses of a perfectly competitive economy to a productivity shock.
However, the recent literature which has emerged independently from the RBC theory has investigated other sources of volatility, especially in the context of developing countries.
These sources of volatility will be discussed in detail in section I.3.
$ )
A necessary condition for measuring volatility in an economic time series is that the series of interest must be ! – meaning its mean and variance should be constant over time. However, many economic variables are !in level; they fluctuate around a changing mean and the size of volatility varies over time. For example, the GDP (Gross Domestic Product) is usually non*stationary, which increases with varying average and its fluctuations around this rising trend are also variable. There are two major ways to make a series stationary.
The first method is to simply take the of the series. Although first*
differencing may not always be sufficient to obtain stationarity; sometimes a second difference may be necessary. First*differencing is, in fact, akin to taking the growth rate of the series. If the variable is expressed in logarithmic form, then first*difference approximates a growth rate, as shown in the following equation:
1
1
log( ) log log log <
< < <
− <
−
= − =
(4)
The second method is to separate the permanent component ( ) from the transitory component ( ! ) in the data. Once the permanent component is removed from the data, the cyclical component can then be analysed. Several methods have so far been proposed in econometrics for decomposing a series into trend*cycle elements. Here, I restrict myself in explaining very briefly the two most widely used methods, namely the Hodrick*
Prescott filter (Hodrick and Prescott, 1997) and the Beveridge*Nelson decomposition (Beveridge and Nelson, 1981).
The 0 * / extracts the trend ( ) by minimizing the following sum of squares program:
{ } 1
1
2 2
1 1
1 2
min & ( ) [( ) ( )]
& &
< λ
=
−
+ −
= =
− + − − −
∑ ∑
(5)where λ is an arbitrary constant reflecting the cost or penalty of incorporating fluctuations into the trend. The first term in expression (5) is the penalty associated with the deviation of the adjusted trend ( ) from the actual series (< ). The second term penalizes the adjusted trend if its growth over a period is very different from its growth in the previous one. Thus λ acts as a smoothing parameter; it controls the smoothness of the adjusted trend ( ); if
λ→0, the trend approximates the actual series. If λ→ ∞, the trend becomes linear. The value ofλ depends on the frequency of data with the standard measures being λ=100 for annual data, λ=1600 for quarterly data, and λ=14400 for monthly data.
The trend component in Hodrick*Prescott decomposition is therefore a weighted average of past, present and future values. The ! component is the residual which is defined as:
=
1 1
1 =
< < < −
=−
= − = −
∑
(6)The second method, 6 # > , states that any unit root process can be written as a sum of a random*walk process and a stationary process:
(1 :) $(1) 7 :( )
≡ − = − (7)
where L is the number of lags; $ :( )= +1 $ :1 +$ :2 2+$ :3 3+... is a polynomial with infinite degree; A(1) can be interpreted as the multiplier of a shock observed in t; and is a
random variable which constitutes a shock. $(1) represents the non*stationary component and ( )7 : is the stationary component.
( ) (1) ( )
7 : =$ −$ : is verified, and by construction C(1)=0.
2 3
1 2 3
2
1 2 3
( ) (1) ( ) (1 ) (1 ) (1 ) ...
( ) (1 )[ (1 ) (1 ) ...
7 : $ $ : $ : $ : $ :
7 : : $ $ : $ : :
= − = − + − + − +
= − + + + + + +
( ) (1 ) ( )
7 : = −: 6 : (8)
where B(L) represents a polynomial of lags. By replacing (8) in equation (7), we obtain:
0
(1) ( ) (1) ( )
1
$ 6 : $ 6 :
: = −
= + = +
−
∑
(9)Equation (9) shows that is composed of a trend component, called
because it depends on the sum of all shocks since the initial date, and a cyclical component which is stationary. We also observe that there is a serial correlation between the stochastic trend and the cyclical component, because they are both affected simultaneously by the same shock .
Once a non*stationary series is made stationary, there are several techniques to measure its volatility. Following are some of the most usual techniques:
Mean absolute deviation: 1
( )
& −
Standard deviation: &1
∑ (
− ( ))
2Coefficient of variation: ( )
( ) 3
Relative standard deviation (in %): ( ) ( ) 100
3 ×
All these measures of volatility are calculated either on the cyclical component of the series (already in log), or on the growth rate (equivalent to the logarithmic first difference in level) of the series. Hence, the standard deviation must be between 0 and 1. The relative standard deviation is expressed in percentage, and it is useful when comparing two or more series with different units or scales.
$ * +
One of the macroeconomic features of developing countries which distinguish them from advanced economies is a higher degree of economic volatility. Empirical studies show that macroeconomic volatility is “negatively” correlated with the level of income of the country. Figure 1.1 shows that developing countries with lower level of income per capita tend to have higher growth volatility, while developed countries with higher income per capita enjoy less volatile growth.
Rand and Tarp (2002) found that output volatility in developing countries is 15 to 20 percent higher than that in developed countries. Developing countries also show considerable persistence in output fluctuations (Agénor et al. 2000). Malik and Temple (2009) observed that over the period of 1960*
1999, “the median (across countries) of the standard deviation of annual growth rates was more than three times higher for low*income countries than for OECD member countries.” The explanations behind this stylised fact are that developing countries have
‘underdeveloped economic structures’ such as underdeveloped financial
sector, weak institutions, weak automatic stabilizers, inadequate and undiversified trade structure, distortionary policies and microeconomic rigidities. These elements will be elaborated in detail in section I.3.
Historically, developed countries have enjoyed stable macroeconomic performance since the 1980s, while, in contrast, macroeconomic volatility has severed in the developing world. Figures 1.2 and 1.3 compare macroeconomic volatility in the United States and in Argentina, respectively. The U.S. economy has become much less volatile since 1985, as the volatility of GDP growth has fallen by more than half since then. Many observers refer to this
? 9 @ A
phenomenon as the “Great Moderation.” Conversely, volatility in the Argentinean economy has increased since 1980, both in magnitude and in frequency; crisis volatility has appeared more often and more severely in Argentina.
Empirical studies have also found that output fluctuations in developing countries are positively correlated with economic activity in industrial countries and negatively correlated with real interest rates in such countries (Agénor et al. 2000; Kouparitsas, 2001). This relationship could be significantly important for those developing countries which have
! "# $%&'(&&)
? : !; A
( *+ "#
? % 6 % B
, - "#
substantial trade links with industrial countries. Agénor et al. (2000) found that for many of developing economies that have positive correlations with the economic activity in advanced economies, “the correlations generally peak at or near a zero lag, suggesting that output fluctuations in industrial economies are transmitted fairly quickly.” Business cycle conditions in industrial economies could also influence fluctuations in developing economies through
the world real interest rate. The latter is likely to have an important effect on economic activity in developing world, not only because it affects domestic interest rates, but also because it reflects credit conditions in international capital markets.
+ . / 0
This is similar to the first stylised fact, but what differs is the explanation given for the source of volatility. The first stylised fact stated that developing countries experience higher volatility because they have underdeveloped economic structures, but in here the argument is that it is simply because they are “economically smaller” (Crucini, 1997; Easterly and Kraay, 2000; Ahmed and Suardi, 2009). Technically speaking, the argument is based on the ‘aggregation’ of idiosyncratic shocks to individuals in an economy (Canning et al., 1998).
The transfer and aggregation of shocks depend on the strength of correlations or interactions between individuals and on the strength of microeconomic links between agents in the economy. At the aggregate level, macroeconomic volatility (as the aggregation of all idiosyncratic shocks at micro level) declines with the size of the economy because the aggregation of shocks is not perfectly linear. In short, the larger the size of the economy is, the smaller the magnitude of volatility will be (Canning et al., 1998).
Crucini (1997) compared G7 countries with 68 smaller countries, using a one*sector two*country general equilibrium model in which the ! source of heterogeneity is the economy size. He showed that even if developing countries were developed, had the same market structure, the same financial, monetary and fiscal institutions, and faced the same underlying disturbances, they would still experience more severe business cycles. He gives the following explanation for this phenomenon: Consider two countries with substantial difference in their size of economy, and suppose that productivity rises in the smaller economy while remaining unchanged in the larger economy. Physical capital will flow from the larger country to the smaller country until the marginal product of capital is equated internationally.1 Owing to the asymmetry in economic size, the total amount of world capital that must be reallocated is quite small since each unit per capita reduction in capital in the large country increases the per capita capital stock in the small country many times over (moving the marginal product of capital in the small country downward very quickly). As a consequence, the changes in investment and output in the small country, in response to both domestic and foreign shocks, will be much larger than those in the larger country. Therefore, internal and external shocks generate more severe fluctuations in small countries.
1 ' '
The relationship between cyclical movements in the terms of trade and output fluctuations has been found to be significant and strong. The greater the openness of the economy, the greater is the correlation between terms*of*trade and growth volatility. This issue will be elaborated in detail in section I.3.1.
2
Establishing stylised facts about the cyclical behaviour of real wages has important implications for discriminating among different classes of models. For instance, Keynesian models imply that real wages are countercyclical, whereas equilibrium models of the business cycle imply that real wages are procyclical. Empirical studies, however, support the fact that there is a procyclical variation in real wages (Agénor et al. 2000; Male, 2009). Real wages increase in periods of expansion and higher growth, and decline in periods of recession and slow economic growth.
1 The argument does not require exact equality of marginal products across countries, but only requires that capital flow should be in the direction which could reduce the difference between the countries.
&
Recent research and studies have found that macroeconomic instability has significant costs in terms of welfare loss, increase in inequality and poverty, and decline in long*term growth. Below are some of the most important consequences of macroeconomic instability as identified by empirical and theoretical studies.
& $ , - . / !
Theoretical growth models, such as the AK and Schumpeterian models, suggest that volatility induces a growth rate (Aghion and Banerjee, 2005). In an AK model, long*
run growth is entirely driven by capital accumulation, and the average growth rate depends positively on the savings rate. Macroeconomic volatility will have ambiguous effects: (i) to the extent that it increases uncertainty about future income, individuals increase precautionary savings, which in turn leads to a higher equilibrium savings rate and thus higher average growth rate; (ii) But to the extent that macroeconomic instability is associated with higher uncertainty about the expected return to saving, it may reduce the propensity to save, thereby lower growth rate. At the end, the dominance of these two opposing effects depends on the intertemporal elasticity of substitution in individual consumption over time. If the intertemporal elasticity of substitution is greater than 1, the final effect of macroeconomic instability is to reduce the expected return to saving and thus discourage savings. But the empirical results show that the intertemporal elasticity of substitution is generally less than 1, and therefore volatility increases the growth rate.
In a Schumpeterian model, growth is generated through short*run capital investments and long*term productivity*enhancing investments such as R&D, and organisational capital.
During the periods of recession, there is lower return to productive capital investments due to lower demand. On the contrary, the opportunity*cost of productivity*enhancing investments is lower. Hence, firms engage in R&D and creation of organisational capital. These productivity*enhancing investments during economic recessions will finally increase the future long*run growth.
Empirical studies, however, have found totally different results. In a seminal paper, Ramey and Ramey (1995) pointed out that volatility is not neutral; it has adverse effects on growth. They showed that countries with higher volatility have lower mean growth, even after controlling for other country*specific growth correlates. They explained that “the negative effect of volatility stems mainly from volatility of innovations to GDP growth, which reflects uncertainty.”
Macroeconomic volatility hampers growth through creating uncertainty in the macroeconomic environment and depressing the private investment. In fact, investment is subject to irreversibility and asymmetric adjustment costs. Following exogenous shocks, private capital formation will be negatively affected (Agénor, 2004) and private investment declines. There are also several other channels through which macroeconomic instability may affect private investment. In the presence of uncertainty in the macroeconomic environment, risk*averse firms will not invest in risky activities and will reallocate resources to safer yet less productive activities. Therefore the level of capital accumulation may decrease in the economy. Macroeconomic instability also affects the “confidence” of economic agents, which can discourage domestic investment and lead to capital flight – which has potential adverse effects on long*term growth. If macroeconomic instability is conjoined with higher level of inflation, it may lower investment by distorting price signals and the information
content of relative price changes (Agénor, 2004). In addition, a high variable inflation rate has adverse effect on expected profitability – if firms are risk averse, their level of investment will fall.
Hnatkovska and Loayza (2005) found that a one*standard*deviation increase in growth volatility leads to 1.3 percentage*point drop in the growth rate – which represents a sizeable loss. Under a crisis situation, the loss would further increase to 2.1 percentage points
3 "# $%&'(&&&
? : !; @ A
% 4 5
? ( C
of per capita growth rate. They also found that the adverse effects of volatility on growth is larger in countries that are poor, institutionally underdeveloped, undergoing intermediate stages of financial development, or are unable to conduct countercyclical fiscal policies.
It can also be argued that if macroeconomic instability affects negatively the long*
term growth, then it may also slow down the development process in the country, since having a sustainable growth is a necessary condition – if not sufficient – for the development.
Did the countries which enjoyed “better” macroeconomic stability developed faster compared to those that suffered from serious macroeconomic instability? The answer to this question is yet to be explored. It can be an important area of research for future studies.
& & # -
Macroeconomic instability can affect poverty through its impact on
@ Cross*country studies have found a negative correlation between volatility and inequality. Figure 1.6 plots the relationship between growth volatility and income inequality (measured by the income share of the bottom quintile) over the period 1957*1999. However, the causality between inequality and volatility can go in both directions. On the one hand, macroeconomic instability can lower incentives for human capital accumulation which is a good determinant of the level of inequality. Volatility affects different segments of the population differently – depending on the nature and source of macro volatility; it may affect negatively the poor while benefiting the rich. In fact, at the trough of a business cycle, since
the poor do not have self*insurance mechanisms they are affected through a reduction in their income. While the rich who are well protected by self*insurance mechanisms may not experience any decline in income. Hence, growth volatility may increase inequality between the rich and the poor. On the other hand, inequality itself can increase social instability and thus macroeconomic volatility.
Laursen and Mahajan (2005), after controlling for the endogeneity between volatility and inequality, found that the negative effect of macroeconomic volatility on income inequality is statistically significant and robust. They also found that the magnitude of this effect is different across regions which may be due to differences in structural characteristics and in risk*management mechanisms.
Macroeconomic instability can affect income distribution through 5 different channels (Laursen and Mahajan, 2005): relative prices between different goods and services or between factor inputs and outputs; labour demand and employment; returns on physical assets and capital gains or losses; public or private transfers; and community environment effects. The relative importance of these different transmission channels depends, however, on the # !. For example, the effect on income distribution of a macroeconomic volatility that is induced by a shock to agricultural commodity prices is different than a one induced by a financial shock. Nonetheless, in most cases the poorest segment of the population bears the largest burden of the adverse effects of macroeconomic instability. First, their income sources are less diversified – usually their only source of income is their labour earnings. Secondly, their lower levels of assets and limited access to financial services make it more difficult to seek self*insurance. And finally, the poor depend more on public transfers and social services, mainly for health and education, which are likely to be cut during the periods of crises.
Hence, by raising income inequality, macroeconomic instability can contribute to an increase in poverty in the society. Negative income shocks may affect income distribution either ! which increases transitory poverty, or ! which in this case exacerbates chronic poverty (Laursen and Mahajan, 2005). Even if effective poverty*
alleviation and pro*poor policies are undertaken in order to halt the impact of macro instability on poverty, it is suspected that macroeconomic instability will “result in slower poverty reduction for a given average rate of growth” (Guillaumont and Korachais, 2008).
As stated earlier, the causality can also run from inequality to macroeconomic instability. A high degree of inequality has not only negative implications for long*term
development but also for short*term macroeconomic fluctuations (Agénor and Montiel, 2008). Income inequality may create social instability which can exacerbate macroeconomic instability. In addition, countries with high levels of income inequality tend to have a small and volatile tax base; this may translate into high volatility of public expenditures. Iyigun and Owen (2004) argued that income inequality may engender private consumption variability when the ability to obtain credit depends on income. Using cross*country panel data for the period 1969*1992, they found that in high*income countries, greater income inequality is associated with more growth volatility in consumption and real GDP, whereas in low*income countries, higher levels of income inequality tend to be associated with less volatility. A possible reason for such different effects in high* and low*income countries is that financial development and availability of credit are positively associated with higher levels of per capita income. Ghiglino and Venditti (2007), using a neo*classical growth model with preference heterogeneity functions, showed that wealth inequality may also lead to endogenous fluctuations in growth. Therefore, developing countries which are characterized by inequality ! experience macroeconomic instability.
& ' 0
Macroeconomic instability has both a direct and an indirect welfare cost for the economy. Its direct welfare loss is generated through causing consumption volatility. Studies show that the welfare gains from reducing consumption volatility can be substantial (Loayza et al., 2007). It also entails an indirect welfare cost through its adverse effect on income growth and development.
Lucas (1987) in his famous book -) 6 7! . tried to estimate the welfare costs of economic fluctuations, as he himself puts it, in order “to get a quantitative idea of the importance of stabilization policy relative to other economic questions.” Lucas estimated that the welfare costs of economic fluctuations are very insignificant; merely 0.05 percent of consumption per capita. A number of recent studies, however, have questioned this finding. Reis (2006) found that the welfare cost of macroeconomic volatility is significantly higher than what Lucas had calculated. Reis estimated that the costs of eliminating the uncertainty that induces macroeconomic volatility are between 0.5 and 5 percent of per capita consumption. He explains that such a significant welfare loss is caused by its impact on precautionary savings and investment. Reis calibrated his model using the U.S. data. In terms
of nominal value, 5 percent of household consumption could represent more than US$ 450 billion,1 which is a substantial cost for the society.
Pallage (2003) argued that the welfare costs of macroeconomic volatility are substantially larger in poor countries than in the United States. Using several models, including Lucas’ (1987), he computed the welfare cost of aggregate fluctuations in LICs and then contrast these costs with estimates obtained from the same models using US data.
Pallage found that the median welfare cost of business cycles in LICs typically range from 10 to 30 times its estimate for the United States. He also emphasized that for poor countries “the welfare gain from eliminating aggregate fluctuations may in fact be so large as to exceed that of receiving an additional 1% of growth forever.” Although Pallage’s estimates cannot be taken as an welfare cost of macroeconomic instability in LICs, what is certain is that its welfare loss is much larger in poor countries than in the advanced economies. In fact, macroeconomic volatility disproportionately affects the poor because consumption patterns are much more sensitive to fluctuations in income at low levels of income.
These recent findings may suggest a re*thinking of economic policies in poor countries. Washington*based international institutions have always recommended developing countries the policies which focused exclusively on generating growth. Yet not many countries succeeded to obtain long*term stable growth. Despite landmark achievements in economic theory, economists have not yet been able to offer an ultimate solution for countries which suffer from growth*retarding characteristics (Easterly, 2002). A fair approach would be to accompany growth*enhancing policies with measures that aim to smooth out economic fluctuations and to bring about macroeconomic stability in the country.
1 According to OECD Statistics, final consumption expenditure of households in the United States was $9,742.5 billion in current USD as of 2009. Five percent of which is $487 billion.
' +
Empirical studies have identified numerous factors which may induce volatility in macroeconomic aggregates. These factors are of different nature, and I classify them henceforth under and factors. By these two terms I do not imply that they are “external” or “internal” factors in regard to the economy, but whether these factors can be controlled by the government and can be influenced by economic policies and structural reforms. I rely on the results of empirical studies in my approach to enlist the sources of volatility. A large part of these studies use econometric models and techniques to identify the causes of macroeconomic volatility. Nonetheless, other studies are based on the calibration of theoretical models (e.g. general equilibrium models, dynamic stochastic models, etc.) which I do not develop in detail, rather I focus directly on their estimation results.
' $ 1
60 ' ' 7
External shocks have significant impact on macroeconomic instability in small open economies. Above all, the terms*of*trade shocks (fluctuations in the relative prices of exports to imports) are believed to be more pronounced because most small developing countries are price takers in international markets. Some of these countries have very low level of domestic production; not only low manufacturer output but also insufficient agricultural production.
They are heavily dependent on imports; on imported capital goods, intermediate inputs, and on primary food and non*food commodities. Therefore, world price shocks affect these countries much severely. Moreover, these countries export only few primary commodities, and rely heavily on their export earnings for the payment of their large foreign debt services.
Their export revenues are also highly unstable due to recurrent and sharp fluctuations in world demand and prices, which make these economies more and more vulnerable. Given such structural characteristics, it is easy to conclude that small open developing countries are much prone to external shocks, especially to shocks in their terms of trade.
Empirical studies have supported the fact that terms*of*trade shocks account for a significant portion of macroeconomic volatility in developing countries. Mendoza (1995) found that terms*of*trade disturbances explain 56 percent of output fluctuations in developing countries. Kose and Riezman (2001) estimate that terms*of*trade shocks account for almost half of the volatility in aggregate output in Africa. Kose (2002) modelled a small open economy under a dynamic stochastic model, and by using a variance decomposition method,
he estimated that “world price shocks account for a significant fraction of business cycle variability in developing countries.”
Broda (2004) discriminated between fixed and floating exchange rate regimes in his study, and concluded that short*run real GDP volatility in response to terms*of*trade shocks is smaller in countries with flexible exchange rate regime (floating) than in those with fixed regime (pegs). He estimated that in developing countries, terms*of*trade disturbances explain 30 percent of real GDP fluctuations in fixed exchange rate regimes compared to 10 percent in flexible exchange rate regimes. Although a few other studies have concluded that the level of impact of external shocks on output volatility might be lower (cf. Raddatz, 2007), nonetheless, there is no doubt that “exogenous volatility spillovers from abroad” are “a relevant determinant of output volatility” (Bandinger, 2010).
The principal transmitting channels of externals shocks are trade and financial integration. Countries more open to the world economy, which lack sufficient domestic production of primary commodities, tend to be more vulnerable to external shocks; given the fact that most of world price fluctuations occur in primary commodities, in both food and non*food (e.g. oil) items. Financial integration, on the other hand, makes countries more prone to global financial shocks, credit restraints, and world interest rate fluctuations. The level of specialisation of a country also plays an important role in determining the impact of external shocks. Countries more diversified, both in their export and production structures, will be able to decrease the negative effects of external shocks. These structural factors which determine the impact of external shocks over an economy will be discussed more in detail in section I.3.2.
The above arguments concerning external or terms*of*trade shocks were in two directions. On the demand side, large “importers” are more vulnerable because they do not have domestically*produced substitutes. And on the supply side, “specialized exporters” are also prone to the fluctuations in world commodity prices because they are price*takers at the global level. A special case in the latter category is the resource rich developing countries.
Countries abundant in natural resources experience large volatilities especially in their fiscal indicators, because a large part of their revenues is based on their commodity exports. In periods of booming prices of commodities (e.g. oil), countries receive large surpluses and rents from their commodity exports. As Collier (2008) explains, in booming periods, they plan large investment projects for the short* and medium*run, and increase their government expenditures. But when the commodity prices fall, there is a sudden drop in fiscal revenue, and the government can no longer continuously finance its projects which are in the course of
implementation. Moreover, once the government increases its expenditures, it cannot easily reduce it back due to political and social constraints. Hence, in the periods of falling commodity prices, the resource*rich developing countries which do not have good fiscal management experience large fluctuations in their fiscal indicators (e.g. enlarging fiscal deficit, increasing tax rates, or decreasing public expenditures). Fiscal fluctuations will also cause volatility in other macroeconomic aggregates via the consumption channel, as households quickly adjust their behaviour to falling wages or decreasing employment.
+ ' 7
Supply*side shocks, such as productivity or climatic shocks, contribute significantly to output volatility in developing countries. Hoffmaister and Roldós (1997) studied macroeconomic volatility in Asian and Latin American countries, and concluded that supply*
side shocks play a substantial role in explaining output volatility “even in the short*run.”
Kose (2002) estimated that productivity shocks explain 10 to 20 percent of sectoral output volatility in small developing countries.
Agriculture*dependent countries which have not yet achieved a agricultural intensification are much vulnerable to climatic shocks. The irrigation system in these countries is not well developed, and their agricultural output is heavily dependent on climate conditions. Modified agricultural seeds which are flood*resistant and drought*tolerant are not widely used among the farmers. Therefore, climatic shocks, such as drought, flood or other natural disasters, have more adverse effect in these countries than in developed economies.
7
Malik and Temple (2009) investigated the volatility effects of market access (proxied by coastal access), geographic predisposition to trade, climate variability, soil conditions, and ecological classifications of tropical location. They found an especially important role for market access: “remote countries are more likely to have undiversified exports and to experience greater volatility in output growth.” In fact, natural barriers to trade (such as being located far from international markets or having costly access to markets, for example, due to being landlocked and not having an easy access to sea) may lead countries to specialize in a narrow range of exports. This could explain the association in the cross*country data between coastal access, export concentration, exposure to world price shocks, and output volatility, as shown in Figure 1.7. Landlocked countries and/or countries with greater coastal distance tend to have more concentrated exports and thus experience higher volatility. Natural*resource
abundance is also associated with export concentration. Countries abundant with and dependent on point*source natural resources (such as fuels, minerals and plantation corps) have higher degree of exportation concentration, hence higher growth volatility.
Malik and Temple’s (2009) argument is that geographic location influences the prices of intermediate inputs faced by domestic producers, and especially the prices of capital goods, due to high transportation costs. Output growth, thus, tends to be more volatile in countries situated in remote geographical areas. This phenomenon was confirmed in an earlier paper by Brunner et al. (2003) who showed that countries with higher trade costs may experience more volatile real exchange rate and volatile output growth. Malik and Temple (2009) also controlled for the countries’ institutions in their regressions and found that “even when conditioning on institutional variables, geographical characteristics continue to play an important role in explaining volatility.”
8 9 9 1 1 9
? ) & D
“A first look at the geography of output volatility. The top*right panel shows the well*known association between volatility and terms*of*trade volatility. Reading the remaining figures clockwise, volatility in the terms of trade is related to export concentration (lower*right) which is related to mean distance from the coast (lower left) and hence mean distance from the coast and output volatility are positively associated (top*
left). The solid line is a least*squares fit, the dashed line a robust (least trimmed squares) fit.” (description by the authors)
+
Rodrik (1999) studied the question that ‘why some economies were hardly affected by the volatility in their external environment during the second half of the 1970s, while others suffered extensively for a decade or more before starting to recover.’ To answer this, he advanced the hypothesis that domestic social conflicts are key to understanding this phenomenon. He emphasized that “social conflicts interact with external shocks on the one hand and the domestic institutions of conflict management on the other.” These interactions play a central role in determining an economy’s response to volatility in the external environment. “When social divisions run deep and the institutions of conflict management are weak, the economic costs of exogenous shocks – such as deteriorations in the terms of trade – are magnified by the distributional conflicts that are triggered.” In fact, social divisions generate uncertainty in the economic environment, and delay the required adjustments to correct the disequilibria created in the economy. Policy*makers who belong to different ethnic groups will not be able to reach an agreement on bringing necessary structural reforms, or to take effective measures to respond to external shocks. Hence, countries which suffer from social divisions experience stronger volatility effects.
In a complementary but independent study, Tornell and Lane (1999) analysed an economy characterised by weak legal*political institutional infrastructure and by
“fractionalization” inside the government elite. They focused on a fiscal process in which powerful groups dynamically interact and maintain discretionary fiscal redistribution to allocate national resources for themselves. “In equilibrium, this leads to slow economic growth and a “voracity effect,” by which a shock, such as terms of trade windfall, perversely generates a more*than*proportionate increase in fiscal redistribution and reduces growth.”
The authors also note that the governments of such countries would respond in the same perverse fashion # in the case of favourable shocks, by increasing more than proportionally fiscal redistribution and investing in inefficient capital projects. They explain that in a society in which non*cooperative powerful groups exist, the “redistributive struggle”
between them will result in a greater share of resources being invested in non*taxable inefficient activities. In fact, when groups have the power to extract fiscal transfers, due to lack of institutional barriers, such redistributional transfers would be invested in shadow sectors in order to protect their profits from arbitrary taxation.
Volatile aid inflows, too, can be a source of macroeconomic volatility in low*income countries (LICs), especially when aid is in the form of budget support rather than project support. Empirical studies have found that aid flows are usually volatile and pro*cyclical (Pallage and Robe, 2001) and such volatile pattern in aid inflows can have significant negative impact on the variability of macroeconomic aggregates through fiscal indicators.
Aid is observed to be more volatile than domestic revenues, and is rarely stabilizing. In fact,
“unpredictable and procyclical aid can heighten the overall macroeconomic instability” (Bulíř and Hamann, 2008). Arellano et al. (2009) argue that aid volatility induces strong fluctuations in consumption, investment and real exchange rates. They explained that even in the absence of aid, large productivity fluctuations typical of aid*dependent countries introduce high volatility in all macroeconomic aggregates. And when the country receives foreign assistance, aid volatility further exacerbates these macroeconomic fluctuations.
The above arguments focus only on # aid flows. Nevertheless, economists have also emphasised that large aid inflows, in general, can have “Dutch disease” effects. Foreign aid is partially spent on nontradable goods, and, as a consequence, domestic prices increase, which leads to a real exchange rate appreciation. In turn, factors of production (including labour) will be re*allocated to the nontradable goods sector, which will result in a decline in the output of tradable sector compared to the output of nontradable sector. Export competitiveness will deteriorate and it will have an adverse effect on growth (Agénor, 2004).
Thus, foreign aid contributes to macroeconomic instability by appreciating the real exchange rate, and enlarging the trade deficit. Furthermore, foreign aid may damage fiscal sustainability of the recipient country, by decreasing the incentives to implement fiscal and tax reforms. It also weakens macroeconomic stability through “shifting political attention at the margin towards the creation of an ‘enabling environment for aid’ which may not be the same thing as enabling environment for sustainable private sector led growth” (DFID, 2004).
Government will focus on effective management and efficient allocation of foreign aid in the country, and will divert its attention from seeking potential sources of long*term stable growth. Hence, foreign aid affects macroeconomic stability by weakening the
which ensure sustainable growth.
' & 1
Classifying the sources of macroeconomic volatility as and
factors is, in a way, “imperfect.” Although various factors explained in the earlier section
